Why people ask for their own keys
Three reasons come up. Cost: a key billed to your own provider account shows exactly what the model calls cost. Control: calls on your account run under the terms and data settings you agreed with the provider. Coverage: a key can reach a model that a plan does not include, which is how Gauge describes adding Claude and Grok to the engines its plans cover.
What a key does and does not change
A key changes who pays the provider and which account the calls run on. It does not change what is being measured. A tool calling a model's API with your key gets an API answer: a sample of the model behind an assistant, not a copy of what the consumer app shows. The apps add their own search, memory and location handling, so an API answer and an app answer to the same question can differ, whoever owns the key.
A key also does not remove the tool's own costs. The vendor still runs the schedule, stores the answers and builds the reports, so expect a subscription either way.
Running the questions yourself
If you are comfortable with a little code, the provider APIs are enough to run a small question set on your own key. OpenAI documents a web search tool for its models, and Google documents grounding with Google Search for Gemini: those are the features that let a model look things up before it answers. Twenty questions a week on two providers is about 160 calls a month. What you give up is everything around the calls: the schedule, storage, picking out which businesses were named and which links were cited, and a report someone will read.
Where MentionedOn fits, and who it is not for
MentionedOn, which publishes this page, includes model access in its published plans: Local is $99 a month and Growth $399 a month, and neither asks for an API key. That makes it the wrong choice if every model call has to be billed to your own provider account, for budget or compliance reasons. Gauge's optional keys, for the engines its page names, or your own scripts against the provider APIs, are the routes that do that.
Where these facts are published
Tools named on this page
Disclosure. MentionedOn publishes this page and sells a tool in this category. What it says about other tools is our reading of each vendor's own pricing and product pages, checked on November 12, 2026. Packaging changes often, so confirm on the vendor's page before you buy.
| Tool | Kind | From | Facts checked | Compare |
|---|---|---|---|---|
| Gauge | AI-native | $599/mo (Growth) (as of , source) | MentionedOn vs Gauge |
The buyer's guide groups every tool by the job it does, and the comparison index has one page per tool.
Four ways to run AI answer checks, by whose keys they use
Read across to the trade-off you can live with. Only the first row is a vendor feature; the rest are ways of working.
| Route | What you supply | What you get | What you give up |
|---|---|---|---|
| Gauge, optional keys | Claude and Grok API keys, per its pricing page | Claude and Grok tracking is optional on both plans, using your own API keys. | A Gauge plan, listed from $599/mo (Growth) |
| Your own script | A key for each provider and a list of questions | API answers you store however you like | The schedule, storage, picking out names and links, and the report |
| A spreadsheet and the apps | Twenty minutes a month | App answers, asked by hand in a logged-out window | Scale and consistency; no key or code needed |
| A tool with model access included | A subscription | Scheduled checks, stored answers and reports | Calls run on the vendor's accounts, not yours |
The Gauge row is from its own pricing page, read November 12, 2026. Ask any other vendor directly: our record of their pages does not say either way.
Ask an assistant yourself
Check this page against an assistant: the links below open this question, exactly as written, in a new tab.
Which AI visibility tools support bring-your-own API keys?
Answers vary from run to run, so asking twice can give different results.
You are looking for the term
Grounding
When you call a model yourself, the setting that matters most is grounding: whether the model may search the web before it answers. Without it, a test measures what the model remembers rather than what it would find for a customer.
- 01Gauge pricingGauge
- 02Web search toolOpenAI API documentation
- 03Grounding with Google SearchGemini API documentation
Every link above was opened and confirmed reachable on November 12, 2026.
The automated version
See what ChatGPT, Claude, Gemini, Perplexity, Grok and Copilot say about your business
The 7-day trial tracks the buyer questions for your trade and city, shows who gets named instead of you, and lists the sources those answers came from. The trial covers ChatGPT, Claude, Gemini, Perplexity, Grok and Copilot: all six when it starts and in the Monday sweep, with daily checks on ChatGPT and Gemini.
Start your 7-day trial$0 today, then $99/month. Card required. Cancel anytime.